Abstract
Human endogenous retroviruses (HERVs) constitute a significant fraction of the human genome and are increasingly recognized for their roles in both physiological and pathological processes. Despite their biological importance, the annotation of HERV elements remains inconsistent across major public databases. In this study, we present a comprehensive comparative analysis of three key HERV annotation resources: DFAM, Human Endogenous Retroviruses Database (HERVd), and RepBase. We systematically examine their content, classification schemes, and postprocessing workflows and assess the concordance of their annotations based on genomic coordinates. Our analysis reveals substantial discrepancies in element counts, genome coverage, and repeat fragmentation strategies, which we trace back to differences in curation methodologies-ranging from DFAM's hidden Markov model-based automated detection to HERVd's semimanual defragmentation. Using refined matching criteria, we demonstrate that up to 93% of HERV records can be reconciled across databases, yet each source still contributes a substantial proportion of unique elements. We highlight the complementary strengths of these resources and provide practical recommendations for their usage in HERV research. Our findings underscore the need for harmonized standards in retroelement annotation and may inform future efforts toward unified and comprehensive HERV cataloging, particularly in light of emerging genome assemblies such as T2T-CHM13.